Identifying Trace Compounds in Complex Mixtures With MS Data-Processing Software

With the ultimate aim of protecting workers, consumers, and the environment, industry regulations are continually expanding with respect to monitoring the volatile organic compounds (VOCs) emitted by a product or process. As an example, the impending Construction Products Regulation (CPR), expected to be adopted by the end of 2010, specifies the need for in-house screening of chemical emissions from many construction products. Implementation of this new regulation means that manufacturers will need to become conversant with VOC analysis.

Small, heatable microchamber equipment has been proposed as a quick and easy means by which to extract the VOCs from a product, followed by thermal desorption (TD)- GC-MS analysis of the chemicals present. TD is a process used to increase the concentration of VOCs extracted from a sample, leading to more sensitive results. Since this method provides very efficient VOC extraction, the GC-MS data yielded are comprehensive yet complex. This complexity means that individual compounds (particularly trace-level) will be lost in the mass of information, making their identification extremely difficult or impossible. In-house interpretation of these complex emission profiles is particularly daunting to a manufacturing company with limited GC-MS experience; data quality and throughput would be significantly compromised.

Taking this into consideration, TargetView™ software (ALMSCO International, Llantrisant, U.K.) was developed with simplicity as a priority. It allows nonspecialists in GC-MS to obtain meaningful data from the complex chemical emission profiles of real-world samples. The GC-MS data file is imported into the software, which then automatically generates a report listing any compounds of interest found.

TargetView is preprogrammed with a user-specified library of compounds to search for, relevant to the application, e.g., it may comprise a list of compounds specified by a particular regulation. The number of compounds is not limited; it simply relates to processing time.

The software applies sophisticated chemometrics to the GC-MS chromatogram in order to obtain results. Using a three-stage process—dynamic background compensation (DBC), spectral deconvolution, and principal component analysis (PCA)—TargetView interrogates the data and highlights any matches between compounds in the sample and those in the library. Even trace-level, coeluting compounds are identifiable via these three steps.

TargetView is compatible with a variety of commercial GC-MS data file formats. When a file is imported, the software creates a second reprocessed file. The original data file is left intact and can be archived or used for additional functions such as total VOC (TVOC) calculations, if required.

TargetView will also offer real benefits to many industries and GC-MS application areas outside product emission testing. Other examples include the observation of regulated allergenic compounds in a perfume, flavors and contaminants in wine and beer, and the detection of chemical warfare agents in critical public locations such as transport hubs and public buildings. An illustration of the application of TargetView in the construction products sector is described, using plasterboard as an example.

A tool for TD-GC-MS

A circular sample (44 mm diam) of plasterboard was placed into a chamber of the Micro-Chamber/Thermal Extractor™ (Markes International Ltd., Llantrisant, U.K.) unit. A purge gas of helium was applied to the chamber, and the temperature maintained at slightly above ambient for 30 min to equilibrate. Subsequently, air exhaust from the chamber was collected into an industry-standard TD tube (Markes International) containing Tenax® TA (Buchem B.V., Apeldoorn, The Netherlands) as the retaining sorbent. The trapped emissions were thermally desorbed on a UNITY 2™ (Markes International) system and analyzed by GC-MS.

Postrun, the GC-MS data file was imported into TargetView for automatic processing. The target library applied comprised a list of compounds incorporated in construction product regulations.

Compound identification

Figure 1 - Postrun report showing 15 of the target compounds identified in the plasterboard sample.

TargetView processing of the total ion chromatogram (TIC) led to the automatic identification of more than 60 regulatory compounds, most of which would have been extremely difficult for even a GC-MS expert to positively identify. A section of the report displaying target compounds detected is shown in Figure 1. Within the software, the results can be sorted by name, retention time, match coefficient, or area count. 


Inside TargetView

DBC: Removing the interference

Figure 2 - GC-MS data for plasterboard sample showing the effect of DBC (top: original trace; bottom: DBC trace).

The initial stage of dynamic background compensation removes any baseline anomalies from the TIC, such as air/water offset and column bleed. This results in a visibly flatter baseline and leads to a significant improvement in spectral purity, particularly for minor components. A comparison of the GC-MS data before and after DBC is shown in Figure 2.



Deconvolution and PCA: Identifying target hits

It is not surprising that the VOC emission profile obtained from plasterboard indicates a complex sample containing a wide dynamic range of compounds. This degree of complexity inevitably gives rise to compound coelution. Coelution makes the identification of some compounds virtually impossible when using standard library search methods. It is for this reason that spectral deconvolution has been incorporated into TargetView. This process assesses every scan/scan set across each peak, allowing each individual mass ion to be assigned to the appropriate compound.

The deconvolved spectra for each compound are then interrogated by the PCA algorithm, which calculates matches between the sample spectra and the target library spectra. Positively identified compounds are given a match coefficient  (≤1). The user can configure the software so that compounds with a match coefficient below a specified value (e.g., 0.9) can be eliminated to enhance confidence in the data.

Interacting with the software

Figure 3 - TargetView GUI showing the DBC TIC (blue trace, upper window) and identification of the target compound trimethyl benzene (lower window).

The user can interact with TargetView during or after processing. The graphical user interface (GUI) displays the DBC-processed TIC in its top window, and the lower window is used to observe any target compounds found. Figure 3 shows an example display. The target compound trimethyl benzene has been selected from the drop-down box (listing all library compounds); the presence of a bar in the lower window confirms its detection in the sample. The match coefficient and retention time of the identified compound are given; a coefficient of 0.985 for trimethyl benzene indicates a very confident match.

Figure 4 - Comparison of the sample spectrum (top) and the library spectrum (bottom).

If required, the spectrum of the compound identified as trimethyl benzene can be displayed with the library spectrum to visualize the strong match (Figure 4).

The HPlot

In Figure 3, the TIC is overlaid with red bars—the HPlot. These red bars represent compounds detected in the sample. The view can be toggled between all compounds found, including the unknowns, or target hits only (“target hits only” has been selected in this instance). The height of the red bars gives an indication of the peak area for the compound, based on the summation of certain ions found. In this way, the relative abundances of the compounds can also be observed.

Figure 5 - Enlarged window of the TIC containing trimethyl benzene (T), showing  coelution with other compounds.

When the area of the TIC containing trimethyl benzene is enlarged (Figure 5), the HPlot (this time configured for all compounds, not just target hits) shows that several other compounds are coeluting within the peak of interest. One of the coeluting compounds is evidently present at a far higher concentration than trimethyl benzene. The spectrum of this peak, prior to TargetView processing, would therefore contain overlapping mass ions from all coeluting components, making it impossible to accurately identify using conventional library search methods.

Additional capabilities

Confirming hits

Figure 6 - Cross-searching the sample spectrum in NIST database confirms trimethyl benzene.

After processing, if additional confirmation is required, the sample spectrum can be cross-searched in a commercial database, e.g., the NIST library (Figure 6). 

Identifying unknowns

When configuring TargetView to show all compounds found (by unchecking the “target hits only” box seen in Figure 3), the identity of any unknown compound is easily obtained. Clicking on the relevant bar displays the compound’s spectrum, which can then be searched against an established database (such as NIST), as above (Figure 6).

Conclusion

TargetView has been shown to enable accurate and automatic identification of target compounds in complex construction product emission profiles such as those from plasterboard. Despite extensive coelution and many of the compounds of interest being at trace levels, the software provides a reliable result. Even GC-MS experts would find traditional library searching for target compounds difficult and time-consuming under these conditions. To manufacturers apprehensive about looming regulations, the software provides a practical and automated tool, allowing them to produce meaningful emission test results quickly and reliably.

Mr. Roberts is International Business Development Manager, ALMSCO International Ltd., Gwaun Elai Medi Science Campus, Llantrisant CF72 8XL, U.K.; tel.: +44 (0) 1443 230935; fax: +44 (0) 1443 231531; e-mail: garethroberts@ALMSCO.com.

Comments